The general topic of this volume is tagging word tokens with lexical semantic information. The first six papers discuss what information should be tagged and how human tagging (hand-tagging) can be performed so as to maximize accuracy. Hand-tagging is important because it can be used to create standard corpora on which to train and evaluate automated methods for tagging lexical semantic information. The next seven papers concentrate on such automated methods. Automatic tagging should prove useful in a variety of areas including machine translation, information retrieval, information extraction, and dialog understanding.

Although the general topic is tagging lexical semantic information, the majority of the papers concentrate on a more specific problem: word sense disambiguation. However, particular word senses often entail general lexical semantic information. For example, if the sense of bank used refers to something that lies along a river, then one knows that it is a natural formation, inanimate, etc. Consider also the verb bounce which in “she bounced a check” is a telic and in “she bounced a ball” is an activity. Such entailments are especially transparent when using a structured lexicon such as WordNet, which is the case for many of the papers in the volume.